the online dqm of besiii€¦ · besiii sub-detectors performance mdc momentumresolution 0.5%@1gev...
TRANSCRIPT
-
THE ONLINE DQM OF BESIII
Xiaobin JiInstitute of High Energy Physics, CAS
July 9, 2018
-
CONTENTS
• Introduction to BEPCII and BESIII
• The overview of BESIII DQM
• Summary
-
BEPCII AND BESIIIBeijing Electron Positron Collider (II)
Linac
Storage ring
BESIII
2004: Start BEPCII upgrade2009 - : BESIII data takingc.m.s. energy: 2.0 – 4.6 GeVLpeak = 1.0 x 1033 /cm2/s
-
BESIII
Sub-detectors Performance
MDC Momentum resolution 0.5%@1GeV
dE/dx resolution 6%
EMC Energy resolution 2.5%@1GeV
Spatial resolution 6 mm
TOF Time resolution
Barrel 80 ps
Endcap 110 (70) ps
MUC 9 layers RPC, 8 layers for endcap
Trigger rate: ~ 4 KHzData rate : ~ 40 MB/s
-
DATA SAMPLES
5t+t- DsDs
2.4 pb-14 points for t mass scan
5x108 2.9 fb-1
40400.5 fb-1
42601.9 fb-1
43600.5 fb-1
6x109
More data will be collected
46000.6 fb-1
4190 – 4280 8 points3.9 fb-1
44201.0 fb-1
41803.0 fb-1
R Scan, 1.3 fb-1(130 points)
-
DATA QUALITY @ BESIII
• BESIII has three level monitoring• DAQ, no or fast data reconstruction, mainly
the hitmap of each sub-system and electronic• DQM, online full reconstruction, higher
level (physics) monitoring
• DQA, offline full reconstruction with updated calibration constants
• Data quality monitoring Monitor the detector and the data, find problems in time
-
DQM – HARDWARE AND PLATFORM
• 1 server for DQM system management and virtual machine hostIBM x3650 M4, 2 Intel Xeon E5-2630 2.3 GHz
• 5 computing nodes for data reconstruction and analysisIBM Flex System x240, Intel Xeon E5-2620 2.0 GHz, total 24 cores / node
• 2 client machines (PC) for results display
• OS: SLC6, SLC5• Language: C++, python, bash
-
DQM COMPONENTS
• DQM Server• DQM Clients• Histogram handling• Event display• Histogram display• Database (MySQL)• Webpage
-
DATA SERVER
• The aim of the data server is to provide data to all DQM clients• It runs on DAQ machine.• Sampling (copy) data from DAQ data flow• Data is transferred to DQM machine through TCP/IP• Star/stop with run (Controlled by DAQ)
-
DQM CLIENT
• Main component of DQM system• Receive data from DQM server• Using offline software to reconstruct and analyze data• Results are stored in ROOT histograms• Publish histograms
-
HISTOGRAM HANDLING
• Modified from ATLAS tdaq, ROOT based• Histogram server: store all histograms• Histogram receiver: receive published histograms by DQM Client and publish to
Histogram server
• Histogram merger: merge histograms from all DQM Clients, and publish merged histograms to Histogram server• Histogram display: display selected histograms to shifter
Histogram receiver
Histogram server merger
Histogram server
Histogram display
-
HISTOGRAM HANDLING
DisplayServermerger
Server 1
Receiver 1
Receiver 2
Receiver 3
Server 2Receiver 4
Receiver 5
• Modified from ATLAS tdaq, ROOT based• Histogram server: store all histograms• Histogram receiver: receive published histograms by DQM Client and publish to
Histogram server
• Histogram merger: merge histograms from all DQM Clients, and publish merged histograms to Histogram server• Histogram display: display selected histograms to shifter
-
DQM JOB CONTROL
• DQM is run in stand-by mode, when the data-taking is started, it runs automatically• DIM (Distributed Information Management System) is used to synchronize all
components
• After a run is finished, all merged histograms are stored in a root file• A script is used to monitor the new generated root file• A separate job will deal with the root file, obtains useful information of the run, and put
them in the database (MySQL)
• Users can check theses information from the webpage
http://dim.web.cern.ch/dim
-
OVERALL STRUCTION
-
SELF-MONITORING
• Ganglia is used to monitor the whole DQM system
• Crashed jobs: restart automatically• Dead jobs: lose response, do not update histogram any more• Run number monitoring• Job statics, how many events have been processed for each job • CPU usage
-
RESULTS DISPLAY
-
SUMMARY
• BESIII DQM is a lightweight online DQM solution
• Using full reconstructed events to monitor the data
• Separate online DAQ and offline software environment as much as possible
• Expandable
• Successful running at BESIII